importer: 81 of 113 project-buzz records skip on http:// URLs

[themightychris]

The new JSON-based importer (apps/api/scripts/import-laddr.ts) drops about
72% of laddr's project-buzz records (81 of 113 on the 2026-05-18 snapshot)
because their URLs are http:// and the v1 ProjectBuzz.url schema
requires https://.

These are mostly 2016–2018-era press links that haven't been republished
under HTTPS. The links are still publicly served on codeforphilly.org as
plain http://.

Options to evaluate:

1. Schema relaxation — change ProjectBuzz.url to z.string().url()
   without the .startsWith('https://') and let the link work as-is. The
   spec at specs/data-model.md#projectbuzz would need a corresponding
   amendment.
2. Importer upgrade — silently rewrite http:// → https:// in the
   importer with a warning. Some destination hosts may now redirect or
   HSTS-upgrade; others may 404. Verifying each link is out of scope.
3. Drop — accept the loss. The skipped buzz items are old and the
   originating publication may already be defunct.

For now the importer logs [project-buzz] legacyId=<n> missing/invalid URL; skipped for each skip, so the loss is visible in dry-run reports.

Surfaced during the laddr-import-via-json plan (PR #N — TBD).